When requested, ZipFile.getEntry looks up entries named name, then name+'/'. By discarding the trailing slash when calculating hash code, we ensure such entries are found at the same hash table entry. This means we jump around less in the hash table, do fewer calculations per lookup, which ends up being a significant speed-up.

See discusssion and patch suggested by Eirik Bjørsnøs here:
https://mail.openjdk.java.net/pipermail/core-libs-dev/2020-April/065849.html